Navigating with an autonomous robot in unexplored and dynamic environments presents immense complications, particularly when it comes to having to deal with moving targets, and unpredictable obstacles. This is particularly true when considering a wide appendix of contemporary BUG-family algorithms that remain arguably useful in static areas but are inherently limited in flexibility and independence. To overcome these shortcomings, we develop the HyperDynamic BUG (HD- BUG) algorithm as a new navigation framework that couples LIDAR-based perception with adaptive detection scaling, and cost-based navigation. The HD-BUG algorithm defines Light Areas (LA) and Shadow Areas (SA) to further improve the ability of obstacle detection and navigation based on visibility. The algorithm dynamically creates candidate movement points and evaluates them according to a cost function that establishes a relationship between distance-to-goal and likelihood-of-collision. HD-BUG also employs vectorized mathematical operations and batch operations for increased efficiency, as well as dynamically alter the velocity based on a complexity metric when choosing path points. Additionally, HD-BUG employs a reactive local behavior together with global path-planning behavior and predictive modelling. HD-BUG demonstrates a flexible, scalable, and real-time approach to navigation in dynamic and unstructured areas that are often filled with clutter, and change over time. The navigational framework created by this architecture is the first step towards improving the mobility in adaptive applications such as autonomous vehicles, and emergency and disaster response systems.
Hasibuzzaman et al. (Fri,) studied this question.